On 28/06/24 15:32, Christoph Hellwig wrote: > On Fri, Jun 28, 2024 at 11:06:54AM +0300, Adrian Hunter wrote: >> On 27/06/24 17:47, Christoph Hellwig wrote: >>> On Thu, Jun 27, 2024 at 04:44:30PM +0300, Adrian Hunter wrote: >>>> Probably just do what the block layer was doing e.g. >>> >>> That was really just a hack due to the lack of error handling in >>> the old blk_queue_* interfaces. If ADMA really doesn't actually have >>> a limitation, please just don't set any max_segment_size at all. >>> >> >> There is a limitation of 64KiB max. (or less with >> SDHCI_QUIRK_BROKEN_ADMA_ZEROLEN_DESC) per descriptor. > > Hmm, I thought the conflusion was that it can be more, which is why > you proposed to increase it if the PAGE_SIZE is >= 64k. And based on > Jon's report at least for his tegra setups it works with 64k. There is a workaround in that case to split to 32KiB chunks
